Hello,
If I might chime in with what we've done in Ferret to deal with
different calendars, applying what's in the CF standards for calendar
types at
http://www.cgd.ucar.edu/cms/eaton/cf-metadata/CF-1.0.html#cal.
That list includes the various "year" definitions that are in Udunits,
plus a 360-day calendar.
*
*As Jonathon says, "year" isn't a fixed quantity. Neither is "month" for
the same reasons. But the application can use that fact as a basis for
comparing data on different calendars, and for computing the date
corresponding to a time step on any defined calendar. Using the length
in seconds (or days) of the year for the particular calendar we can
convert a time step to the correct date. The definition of the time axis
axis, having for instance, calendar attribute time:calendar = "julian"
and units attribute time:units="days since 1990-1-1 0:0:0" gives us
everything needed to convert the time step 1000 to the correct date in
September 1992.
If model data computed on a 360-day calendar is compared with
observational data on the Gregorian calendar we have a way to make the
appropriate conversion of data on one calendar to the other. Even if
the time unit is "days" or "seconds" we need the information about the
length of the year to convert to formatted dates or to compare data that
has different time axes.
Ansley Manke
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <
http://mailman.cgd.ucar.edu/pipermail/cf-metadata/attachments/20070330/958b8f8d/attachment-0002.html>
Received on Fri Mar 30 2007 - 12:41:32 BST
